Testing Deconvolution Algorithms

Deconvolution is a standard technique in image processing for removing out of field data from fluorescence images (for a nice review see this iBiology lecture by David Agard). We have various deconvolution algorithms available on the workstations at the NIC, which can be confusing for users. So, I tested a variety of them on a standard specimen (a BPAE cell stained with Rhodamine-Phalloidin and an AlexaFluor488 labeled anti-tubulin antibody).


As you can see, the Landweber algorithm seemed to handle this sample well, and there wasn’t much difference between running the algorithm for 10 or 20 iterations (mainly differences in the chromatin). The Blind algorithm appears second best, and Richardson-Lucy and Automatic also fared well. The Fast method was the clear loser. Also, this cell was imaged with our C2 confocal, which is capable of producing such amazingly sharp images. We’re happy to show you how.